The Development of an Energy Efficient Temperature Controller for Residential Use and Its Generalization Based on LSTM

• Published in: Sensors (Basel, Switzerland) Vol. 23; no. 1; p. 453
• Main Authors: Alexandru, Tudor George, Alexandru, Adriana, Popescu, Florin Dumitru, Andraș, Andrei
• Format: Journal Article
• Language: English
• Published: Switzerland MDPI AG 01.01.2023; MDPI
• Subjects: acquisition; Artificial intelligence; Automation; Climate change; Control; Control methods; Controllers; efficiency; Energy consumption; Heat transfer; HVAC; Laboratories; LSTM; Machine learning; Neural networks; Physics; PID controller; Radiation; Residential buildings; simulation; Technology application; Thermal analysis; Thermostat; Virtual environments; Romania; PID controller; efficiency; control; LSTM; acquisition; simulation
• ISSN: 1424-8220; 1424-8220
• DOI: 10.3390/s23010453

Thermostats operate alongside intelligent home automation systems for ensuring both the comfort of the occupants as well as the responsible use of energy. The effectiveness of such solutions relies on the ability of the adopted control methodology to respond to changes in the surrounding environment. In this regard, process disturbances such as severe wind or fluctuating ambient temperatures must be taken into account. The present paper proposes a new approach for estimating the heat transfer of residential buildings by employing a lumped parameter thermal analysis model. Various control strategies are adopted and tuned into a virtual environment. The knowledge gained is generalized by means of a long short-term memory (LSTM) neural network. Laboratory scale experiments are provided to prove the given concepts. The results achieved highlight the efficiency of the implemented temperature controller in terms of overshoot and energy consumption.